Comfy Motion Track Control
Generate local LTX 2.3 videos guided by the configured HDR IC-LoRA through the motion-track control workflow. Use when the user wants an image animated along drawn sparse motion trajectories, spline overlays, point tracks, or a prepared motion-control reference video. Do not use for style-only LoRAs, Seedance API video, image-only generation, music-only generation, automatic point tracking from r…
Comfy Media
Review generated comfy-agent-tools media, build local indexes, serve the gallery, and export selected artifacts into a HyperFrames review-reel project. Use after comfy-imagegen, comfy-videogen, or comfy-musicgen creates outputs, or when the user wants to browse, compare, select, or compose generated media.
Comfy Musicgen
Generate music locally with comfy-diffusion and ACE-Step 1.5 Base. Use when the user wants local GPU-backed music or song generation saved as WAV in the workspace. Do not use for hosted audio APIs, image generation, video generation, voice cloning, speech-only TTS, model downloads, ComfyUI server workflows, UI work, or custom node installation.
Comfy Tools Setup
Bootstrap and validate the comfy-agent-tools Python CLIs for agent use. Use when the user asks to setup, install, update, or diagnose comfy-agent-tools; when a required CLI such as comfy-imagegen, comfy-imagedescribe, comfy-videogen, comfy-musicgen, or comfy-models is missing; or before another comfy skill runs a CLI on a new machine.
Comfy Bernini Videoedit
Run Bernini WAN 2.2 video edit workflows with comfy-videogen. Use when the user asks for Bernini video editing, video-to-video V2V, reference-video-to-video RV2V, video-reference-to-video VV2V, or reference-to-video R2V using the local wan22-bernini profile. Do not use for generic WAN generation, Seedance remote API video, image-only generation, model onboarding, or unsupported custom ComfyUI ser…
Comfy Lora Onboarding
Help organize and select local ComfyUI LoRAs for comfy-agent-tools. Use when the user wants to add, move, rename, organize, inspect, or apply LoRAs; when a requested LoRA is ambiguous; or when LoRAs are loose in /mnt/models/comfyui/loras and should be arranged by architecture and purpose.
Comfy Imagedescribe
Describe or caption local images with the Qwen3-VL 2B Instruct vision-language model through transformers. Use when the user wants a local GPU-backed image description, caption, tagging, or VLM question-answering over an image file. Do not use for image generation, editing, upscaling, video generation, music generation, model downloads, ComfyUI server workflows, UI work, or custom node installati…
Comfy Model Downloader
Download missing built-in comfy-agent-tools model files on demand by capability. Use when a generation/edit/upscale/music/video request needs local models that are missing, when comfy-models reports missing_model_file, or when the user asks to download supported base models. Do not use to download all models unless explicitly requested.
Comfy Imagegen
Generate, edit, or upscale raster images with comfy-diffusion, including local Anima Base v1.0 with turbo LoRA, Qwen Image Edit 2511, FLUX.2 Klein 9B SNOFS, local Ideogram 4 structured prompt/bbox generation, local Krea2 Turbo, ClearReality, and remote Grok Imagine API nodes. Use when the user wants image generation or image editing from the current machine with outputs saved into the workspace.…
Comfy Model Onboarding
Configure local comfy-agent-tools model profiles and defaults. Use when the user has no .comfy-agent-tools.json config, wants to set models_dir, change a default capability profile, add a new checkpoint or fine-tune of a supported architecture, or diagnose profile/config errors such as unknown_profile, unsupported_capability, architecture_mismatch, config_error, missing_model_file, or unsupported…
Comfy Videogen
Generate or post-process MP4 videos with comfy-diffusion using local LTX 2.3, local WAN 2.2, NVIDIA RTX Video Super Resolution, local SeedVR2 video upscaling, or remote ByteDance Seedance 2.0 API nodes. Use when the user wants local GPU-backed text-to-video, image-to-video, image+audio-to-video, video+audio processing, first/last-frame video generation, WAN 2.2 image/first-last-frame/sound-to-vid…